Statistical Methods
Choose Sampled-Curve Quadrature with the Grid in View
Define the task
Sampled-curve integration depends on what is assumed between observations. Preserve actual coordinates, their units and their order; do not silently replace irregular times with equally spaced indices. Simpson quadrature uses local polynomial information, whereas a trapezoidal calculation joins adjacent samples with straight lines. Neither repairs measurement error, sparse coverage or an unidentified discontinuity.
Carry out the check
Inspect the sample count and spacing. With an odd number of equally spaced samples, Simpson integration is exact for polynomials through degree three; unequal spacing reduces that stated exactness to degree two. Verify how your software handles other grid patterns, and compare against a known function before applying a workflow. A finer-grid comparison can reveal sensitivity, but agreement between two numerical rules does not prove that an unobserved trajectory is correct.
Worked hypothetical example
Hypothetical example: y=x² at x=[0,1,2] gives y=[0,1,4]. Simpson's rule yields (1/3)×(0+4×1+4)=8/3, matching the analytic integral from 0 to 2. Trapezoids give 0.5+2.5=3. The difference here reflects curvature, not evidence that every empirical dataset should use Simpson's rule. Report grid coordinates, integration limits, units and the chosen rule. Do not attach a sampling confidence interval without a separate uncertainty model.
Method checked 22 November 2026 against Primary method documentation: simpson.